Eurobot-2026-Vision

This repository contains the vision system for the Eurobot 2026 competition.

Start with Docker

Build image and container

On x86 computer:

cd docker/
docker compose up -d

On nvidia jetson computer:

cd docker/
docker compose -f compose.jetson.yaml up -d 

Build workspace

Attach to the running container:

docker exec -it vision-ws bash

Inside the container, build the workspace:

source /opt/ros/humble/setup.bash
colcon build
source install/setup.bash

Realsense

Launch realsense node

Make sure the RealSense camera is connected via USB to your computer, then run:

ros2 launch realsense2_camera rs_launch.py

If you want to launch three camera at the same time, please run:

ros2 launch realsense2_camera rs_triple_camera_launch.py

Aruco

Run aruco detection for robot

ros2 run aruco_robot robot_detector

Run aruco detection for sima

ros2 run aruco_sima sima_detector

GUI

Run rivz

Open RViz to visualize camera images and TF frames:

rviz2

Yolo

Run hazelnut detection

Open RViz too

ros2 launch yolo hazelnut_detect_launch.py

Set localization mode

modes:["bb", "obb_dep", "obb_pnp", "obb_mix"]

ros2 param set /yolo_node mode "obb_mix"

Save Picture

Save RGB pictures for yolo training datasets

ros2 run yolo save_aruco_training_data.py

Save Depth pictures for yolo training datasets

ros2 run yolo save_depth_training_data.py